Message ID | 1512689298-9458-1-git-send-email-arbab@linux.vnet.ibm.com |
---|---|
State | Accepted |
Headers | show |
Series | hdata/iohub: Use only wildcard slots for pluggables | expand |
On Thu, Dec 07, 2017 at 05:28:16PM -0600, Reza Arbab wrote: >We don't want to cause a VID:DID check against pluggable devices, as >they may use multiple devids. > >Narrow the condition under which VID:DID is listed in the dt, so that >we'll end up creating a wildcard slot for these instead. Ping! Could this go in? It resolves an issue for us where the GPU-to-NVlink PCI bindings aren't created in systems with ES parts, which have a slightly different ID than QS parts, eg. 10de:1dbd instead of 10de:1db1.
Reza Arbab <arbab@linux.vnet.ibm.com> writes: > On Thu, Dec 07, 2017 at 05:28:16PM -0600, Reza Arbab wrote: >>We don't want to cause a VID:DID check against pluggable devices, as >>they may use multiple devids. >> >>Narrow the condition under which VID:DID is listed in the dt, so that >>we'll end up creating a wildcard slot for these instead. > > Ping! Could this go in? > > It resolves an issue for us where the GPU-to-NVlink PCI bindings aren't > created in systems with ES parts, which have a slightly different ID > than QS parts, eg. 10de:1dbd instead of 10de:1db1. It can! Merged to master as of 7a1f035b57ab83ec980c487a03c333e2bca8362c and made it into -rc4.
diff --git a/hdata/iohub.c b/hdata/iohub.c index a5f12fd..2260f60 100644 --- a/hdata/iohub.c +++ b/hdata/iohub.c @@ -705,7 +705,7 @@ static void parse_one_slot(const struct slot_map_entry *entry, st_name(type), vid, did); /* The VID:DID is only meaningful for builtins and switches */ - if (vid && did) { + if (type == st_sw_upstream && vid && did) { node = dt_new_2addr(parent, st_name(type), vid, did); dt_add_property_cells(node, "reg", vid, did); } else {